Block-Based Optimization for Frequency-Selective One-Bit Quantization

One-bit quantization converts real-valued signals into binary sequences, offering significant hardware simplification advantageous for low-power and real-time applications. However, it introduces considerable quantization noise, challenging signal quality, especially for signals with spectral complexity. Traditional methods like Sigma-Delta Quantization (SDQ) effectively mitigate quantization noise for lowpass signals by pushing it outside the signal bandwidth, but struggle with more complex spectral requirements.We present a block-based optimization framework (OBBQ) for one-bit quantization, which divides signals into smaller blocks, and optimizes each block independently. By formulating quantization as an optimization problem, our method enhances signal reconstruction accuracy and enables precise noise shaping adaptable to complex spectral content. It outperforms traditional SDQ methods in both flexibility and fidelity, facilitating real-time processing across various desired frequency domains.